Borg KI und das Ende des Interfaces

Transcript

00:00:00Welcome to Think Different, Think AI, the podcast by Mark and Jens.

00:00:07Two technology-loving minds who not only talk about artificial intelligence but live it.

00:00:14Here you'll find clear classifications, genuine practical insights, and a fresh perspective on what is possible.

00:00:20Understandable, critical, and always with a wink.

00:00:24Food for thought, to smile about, and above all, to engage in conversation.

00:00:43Cuters are born. Resistance is futile. Your life as it has been is over. From this time forward you will service us.

00:01:01So, before we have any more iconic Borg moments that send chills down my spine,

00:01:13run down my spine.

00:01:14For those who might not know, that was...

00:01:20Is it even possible?

00:01:21Is it?

00:01:22A short quote from a Star Trek episode.

00:01:25Star Trek, roughly always about space ships, no matter in which episodes, that flew around somewhere

00:01:29through space,

00:01:31with a crew that was more or less cobbled together from all sorts of people, you know?

00:01:37Beings who always tried to carry a bit of humanity's good into the world.

00:01:46And in doing so, they collectively failed.

00:01:48And in this situation, they encountered another collective, the Borg.

00:01:55And the one we just heard, that was John Luke Packer, actually one of the

00:01:59iconic captains of the USS Enterprise that keeps appearing in Star Trek.

00:02:051701 D, I guess that needs to be mentioned as well.

00:02:09Mhm.

00:02:10For the nerds out there, to whom my colleague apparently belongs.

00:02:13And this dear John Luke Packer was assimilated by the Borg.

00:02:20The Borg are a computer collective, a cyborg collective that has invaded all sorts of worlds

00:02:31and assimilated the beings of those worlds.

00:02:35And channeled the cultural and technological achievements of their culture.

00:02:41And we just heard Captain Pika, who...

00:02:45Just assimilated.

00:02:46...now wants to assimilate his colleagues quasi as well.

00:02:51And now we have already talked quite a bit about...

00:02:54Yes, you said that I would make you a single Borg, it's really great that you have already taken away so much.

00:02:59Just briefly, because you're saying that's really a nerd.

00:03:02So yes, at that point I am a nerd, I used to be young back then, it was Star Trek on TV.

00:03:08And there was a talk show on ZDF back then, where you were asked, how much is the tea at James Tea.

00:03:16Kirk. And the solution was louder than the tins when he is the bear. Yes, exactly. And I called in as

00:03:23a teenager and caused havoc on the phone, how you can actually tell such nonsense

00:03:27for this captain, who sucked in his belly calmly here in 1870 to the

00:03:34To move through the endless expanses. I’m still holding in my stomach today,

00:03:38but that's enough about being a nerd. The episode we actually want to record today and

00:03:45want to record, is a bit of a,

00:03:48let’s phrase it like the title, I’m setting the title now,

00:03:52Mark, unplanned, Box, AI, and the end of the interface.

00:03:58Oh, yes, I mean, as we prepared for the episode,

00:04:02you can hardly believe it, there are actually preparations,

00:04:05the idea came up a bit that when we talk about this whole

00:04:08identity thing, we give systems, give tasks, provide tools,

00:04:14that systems also communicate with other systems, and we

00:04:18looked back at what we have actually learned from the

00:04:21Learned from the science fiction of our youth and taken with us, they just

00:04:25nonetheless the topic of, well, if not digital objects that

00:04:31interact with each other to achieve a greater goal, in the

00:04:34science fiction language, we quickly end up with the Borg. The Borg, for

00:04:37those who don’t know, I would like to explain a bit more

00:04:39They are basically living beings that become connected through cybernetic forced implants, you can’t really

00:04:46say that they go to the Apple Store and say, I would like a brain interface.

00:04:49Yes, assimilating is not really the positive way one might picture it.

00:04:55As I said, you don't go into the store saying, I would like a brain interface,

00:04:59yes, there's nothing available for purchase, it might not be available for

00:05:02purchase anytime soon, but you actually get more like forcibly integrated into this community

00:05:06And through that, you develop Kubernetes implants.

00:05:10Some things in your body are essentially replaced.

00:05:13And you lose control over your body, over your actions.

00:05:16And in your head, you hear the voice of all,

00:05:20controlled by Queen Debork Green,

00:05:24who essentially unites the whole collective under her.

00:05:27And basically, the final direction fades,

00:05:30where the collective is supposed to move.

00:05:32In the end, there was a brief meaning,

00:05:34there was later another insurance with TK,

00:05:36they eventually caught the queen on it one time.

00:05:39But other rooms, we might not want to spoil too much of the entire film.

00:05:42Let me also briefly interject.

00:05:44Okay, okay, okay.

00:05:45It's nice, it’s said with that, one doesn’t go in the name and says, I want

00:05:48a brain interface.

00:05:49So for the digital marketers among us, there was also no opt-out.

00:05:53Well, and it was simply, so it didn’t exist for me, I can mix in now

00:05:57take it away again, I don’t want to be part of it, well, I changed my mind,

00:06:01there were such things with the collective.

00:06:02And everyone was equally entitled, right?

00:06:04Everyone was treated according to the motto, dog, cat, mouse, babies, adults, men, women,

00:06:08regardless of you, with the nanotubes and then off to the party.

00:06:13Yes, but for that you also had

00:06:15a shared collective consciousness.

00:06:18So, it is basically the case that

00:06:21the individual drones,

00:06:23agents, you could say today from our interrogated perspective,

00:06:27these individual cyborgs were constantly aware of each other

00:06:32and all felt together, acted together, and the knowledge of the individual

00:06:38agent was the same as the other agents.

00:06:41That was where you also had film moments when you faced a problem,

00:06:46adapted to it, for example in this case just a phaser beam and shields

00:06:51and stuff like that.

00:06:52But following the motto, if I wiped out two Borg with the phaser, then

00:06:57it was almost not so easy for me with the third, fourth, perhaps even impossible,

00:07:01because it simply adjusted to the way

00:07:04I interacted with it, let's say.

00:07:07Yes, yes.

00:07:09What is a bit like now,

00:07:11before we continue to dive deeper into this,

00:07:15if we compare this with the world we have today,

00:07:19when we look at the topic of AI,

00:07:22maybe also first the individual model, the separate model,

00:07:26that is basically still something,

00:07:28when we see that we all use it,

00:07:30this one model, just like GPT appeared, even without current internet knowledge, but

00:07:37only with the knowledge of the respective training. To which we as humans simultaneously

00:07:42accessed, then it was actually a bit comparable to a

00:07:46collective pool of knowledge, which was initially the same for everyone, adaptively responded,

00:07:53depending on which questions I posed, which prompts were input. That is

00:07:56already a bit of a first analogy that has been there since the very beginning of the

00:08:01Jegebiti moments. We suddenly had a sort of collective,

00:08:04not conscious, but a collective shared understanding of

00:08:09topics. To which everyone had democratic access to

00:08:13Jegebiti, also in the same way, to which they could access

00:08:18and can. And there are certainly some people who reach it faster

00:08:24and simple a result, because they master the topic of prompting and the context.

00:08:28gave.

00:08:29But yes, as you say, actually we all had access to this knowledge connectivity.

00:08:35Come on, not just knowledge, you just specified that nicely, actually again

00:08:38something, but actually also ability.

00:08:41So this means, regardless, let's say initially of my, as I said, of course, if

00:08:45I have no idea about this topic, then I will also prompt poorly,

00:08:48whether these are knowledge inquiries or also coding inquiries, but despite

00:08:51all that with the democratization, just in that moment everyone theoretically had the same

00:08:58programming ability.

00:08:59For example.

00:09:00So comparable to the FESA shooting, we've learned to write all Python suddenly

00:09:05at this moment.

00:09:06We weren't all aware of it, but we were actually that far along.

00:09:09You also have to say, it's not just that you have access to this knowledge,

00:09:13because okay, there are different, let's say, pricing models from TechGPT, but if

00:09:18I compare it with what I have to pay in fees to book books,

00:09:24courses, seminars, to engage with people, to take the space and leave

00:09:30and have to go somewhere else for perhaps a training, a conversation with an

00:09:36expert, this really brings everything into the same space

00:09:41and time, that you have everything available at the same time, either

00:09:47in a free model or that cost 29 euros, 229 euros, depending on how the model

00:09:51you signed up for.

00:09:55You invest once and essentially have access to this knowledge base for this month,

00:09:59to these possibilities that the models have also

00:10:02accumulated by now.

00:10:07If we think about it historically, this is definitely one of the first

00:10:13pivotal moments, if you will. Or would you say, the emergence

00:10:17of the GPT moment, whatever we want to call it, of course there are all the other things that were there before

00:10:20and now there are other models too, but let's focus on this moment, because it was indeed

00:10:25a groundbreaking moment, an incredibly democratized access to knowledge, an incredible

00:10:30democratized access to the ability to create something, whether that's, let's take the

00:10:34whole Gen-AI topic around image generation, everything that emerged in parallel during this period.

00:10:39This was definitely a turning point. Also, one could see the internet, the networking of things, as another

00:10:46Moment of enthusiasm, Google maybe as this at least accessing knowledge back then, as

00:10:51it was already there, could already be termed that, going further back, probably next to

00:10:56of course such infrastructure, traffic, where we could also network more strongly,

00:11:01because the world had to grow bigger.

00:11:03The topic then is printing and libraries, where knowledge was basically democratized

00:11:08and also made accessible to a larger amount, then maybe

00:11:14the very first moment of knowledge sharing.

00:11:15But have we really traveled far back in time?

00:11:18When everyone starts with tent activities, I thought we wouldn’t go towards Gutenberg.

00:11:22Then I thought, okay, we’re getting into the topic, okay, then you could use tools like

00:11:26that model doesn't learn from training, but instead can

00:11:31retrieve back later on, then we come to MCP, universal interfaces,

00:11:35allowing our little chatbot to not just write answers for me anymore, but

00:11:42I don’t know, suddenly creates 3D models in Blender because I’m chatting with it. And thus develops

00:11:48this capability, which we initially understood as democratization of knowledge, through

00:11:53this world knowledge at your fingertips, in such a world logic, because you literally

00:11:59didn’t have any idea of Blender before, by the way, still don't.

00:12:02You have to ask Blender to do things. Isn’t that perfect?

00:12:06Well, but now I can do more. Yes, what we, wherever it is said, with the motto

00:12:12that all these KG systems destroy creativity, I would say now.

00:12:16Well, certainly this is naturally difficult for someone who earns their money in visual arts.

00:12:22It’s hard when they start with such a Sora 2 model and say, Anime, movies, no problem, I’ll do it.

00:12:27To be honest, I feel that cheap digital photography has probably led to a boost in creativity.

00:12:36Because more people are allowed to take lots of really bad photos and can express themselves without having to pay for film development or anything else.

00:12:44Every technological advancement simply changes the game.

00:12:49Before photographers, there were only portrait artists, who were also

00:12:56back into the past, my friend.

00:12:58Yes, I just wanted to, because you are currently trying this sensory

00:13:03Moment I just had, so as probably the very first person, to bring the box into relationship

00:13:07with Gutenberg and the printing press.

00:13:09Okay, I'm of course already following.

00:13:11Then I have to somehow bridge back to this sensation.

00:13:15If we stay in the digital realm for a moment, it is indeed the case that this aside from the networking that has previously arisen through the internet, this GPT moment was probably the first Borg moment where we can say we suddenly have something like collective knowledge, collective capability at your fingertips, at your voice, whatever, available that offers entirely new possibilities.

00:13:39So first of all, this positive side of the Borg collective, that everyone in this Borg collective,

00:13:45as soon as the information is available, as soon as the ability is available,

00:13:49can basically fend off the Phaser decision with a shield, so that this is basically now actually

00:13:55done through models, especially downloading open source in the garage on the phone.

00:14:00There is the box of Pandora, which is already opened in the direction of the Borg collective.

00:14:07At this point, I also want to continue my story, saying that,

00:14:11not only do you have this knowledge, but also this tooling, while the box starts

00:14:16and building things and developing their cow bus, you have this topic with us

00:14:23regarding the tools. This includes controlling a 3D printer to making Blender and,

00:14:29and the entire chain, we create an object and print the object, we create the object.

00:14:34That today you have the tools right at your hand and thanks to the internet you can also control them worldwide

00:14:40.

00:14:41And according to that motto, here in the computer, I'm starting a chat and it wouldn't be so

00:14:45hard to say, I'm controlling, what do I know, the 3D printer at my neighbor's

00:14:50around the corner.

00:14:51This is now all possible worldwide in a sense.

00:14:54Right, correct.

00:14:55While of course that was also, let's say, what's now described

00:14:59So it's not really this AI topic, that I say I can control endpoints somewhere.

00:15:08I can literally use technological commands that I would have otherwise carried from village to village

00:15:13in the past, to actually access the 3D printer, the real printer,

00:15:19that drive somewhere in a server room around the world for my data.

00:15:24What struck me a bit like a revelation, we recently had on the site

00:15:28maybe a little later fuzz tracked from the side, a little fuzz tracked. We had

00:15:31an episode that never saw the light of day. Some who know the left episodes know that we

00:15:36recently did a non-recorded live episode at an internal event,

00:15:42where we talked about OpenAI, what has happened in terms of e-commerce, AI as a customer, but we also

00:15:47talked about what OpenAI introduced, namely the whole topic of Etiquette,

00:15:52ChatKit, apps in Open AI. This is what I said stood out because

00:16:00so far we had world knowledge, now we also get world logic, because you no longer

00:16:03leave the chat and can do things, for example, providing an app, so

00:16:09you call them apps, I would now say these are not apps, that's a different topic. They

00:16:13bring something into this chat so you can execute logic in the chat, which

00:16:17again can do something somewhere in the world or at your place. From that side yes, what you say,

00:16:23is of course correct. Things like remote screen controls, data transmission,

00:16:27are nothing new. But with the next steps of Open AI now,

00:16:32you not only have knowledge but also logic at your fingertips without spatial limitations,

00:16:38because you can access companies, services like travel bookings, or whatever,

00:16:45that's not all just app-based, it's all pretty much cranked out, where again tools

00:16:50are interconnected, logic is interconnected, processes are interconnected, can be initiated

00:16:54and in quotation marks you only have this, I’ll write a command in Churchivity.

00:17:01Yes, okay, tooling, so capabilities, not just in the virtual world, but also

00:17:09in the real world, the box has shown, they’ve always been flying around with some

00:17:12Bockcube, which means basically not just, not just the,

00:17:17then I have to make sure not to offend the nerds or say something wrong.

00:17:21They definitely haven't assimilated a star designer who has created a nice spaceship,

00:17:26they are just going at the cube.

00:17:28Cube, now I've only developed the Rubik's Cube.

00:17:31Yes, I mean, the cube is, I mean, we are also, let's say it this way,

00:17:35ship containers are somehow angular and such, so that’s already fine.

00:17:40Previously to choose, Bauhaus. The Powers architecture also talked about it being very

00:17:46frublik in many places. From a design perspective, it's not about the

00:17:50aesthetics, but from a design perspective, it is very efficient. And so it is, the collective,

00:17:54whether it's a bee state or something else, the collective also tries to be very

00:17:59efficient. So accordingly, that’s already fine. Now it is the case that we now

00:18:05And you just mentioned the topic of agency and stuff, networks.

00:18:09We are of course coming to this topic of knowledge and abilities.

00:18:13Currently, are we also getting into a situation where AIs talk to each other.

00:18:19We have already discussed this urgency or other instances in a previous episode.

00:18:23And this is also a thing.

00:18:25So this communication between machines.

00:18:28That machines teach each other something,

00:18:32like the Borg Collective taught themselves things, one learns something somewhere,

00:18:36the others pick that up, in a certain Borg node. I think they were also somewhat like

00:18:41nodes organized, because there is of course still, let's say, spacetime, and the Borgs then

00:18:47were scattered somewhere in the universe and thus information has to flow

00:18:51through the universe in order to enlighten the collective. This means that there were indeed

00:18:56different levels of knowledge in the situation, just probably like in a

00:19:00Borg cube; if I remember correctly, it was not automatically that when the phaser fire

00:19:05on one side stopped working, it was at the same moment on the other side, but

00:19:09it naturally takes time for the information to flow through the Borg cube

00:19:13and much later maybe the other side could intercept photon torpedoes

00:19:17or whatever else existed in Star Trek. This means,

00:19:21that is also a very interesting aspect. This topic of communication in these

00:19:26magnetic networks, when AIs talk to them, I would also like to perhaps

00:19:30discuss this point a bit more, as I am interested in your opinion,

00:19:36then now actually counts, when I have a junction that veers off in the wrong

00:19:38direction and weighs false information more heavily than it actually should be correct,

00:19:45and it becomes harmful to the system.

00:19:47Well, due to harm to the system, another topic comes to mind.

00:19:52One.

00:19:53Hopefully back again, I'm trying not to give the typical politician's speech, no matter what you

00:19:57want from me, I'm selling you a washing machine. But still, I want to get this thought out.

00:20:02The topic of digital sovereignty. I mean, most people think about this topic

00:20:07Digital sovereignty, perhaps because America is currently not interested in providing us with

00:20:13any services as they would like in the future. That’s one thing,

00:20:17but the other thing that I find even more interesting is, because I do think,

00:20:21because I do think that one definitely wants to make money, the question is how territorial

00:20:28do models behave? So how do they weigh things? We had a last topic that

00:20:32you also mentioned to me, when they have the world knowledge, it will be difficult and eventually

00:20:35with a jailbreak, everything can be opened again. But if you use things normally

00:20:39and say, how do you behave, how should you react, then the responses

00:20:43are American models behave differently than Asian models. And I find that the

00:20:47question is, how will that change in the future? Will the gap get bigger or smaller?

00:20:52I believe I have already mentioned, if so, please feel free to leave a scolding in the comments for

00:20:57me, that when you give systems historical events and ask them after a

00:21:02reassessment of the situation, that is, historical political events, the

00:21:07systems already today generally react more and more radical and suggest more radical proposals

00:21:13to solve historical situations, so radical meaning an increase in military to solve

00:21:20historical events. But of course, the point is, as in a whole interconnected

00:21:27world, if a Borg node or a real data flow behaves harmfully towards me, it is

00:21:35the whole topic of cybersecurity, how do we deal with it, how do we notice it, how can

00:21:40Recognizing consequential damages before larger failures or greater

00:21:45influences occur is certainly a point that one needs to discuss,

00:21:50especially as systems become more and more autonomous and interact

00:21:55with each other increasingly autonomously, that through such things,

00:21:59like Prompt Injection by a person, who gets bored if you think about it,

00:22:03that in such a construct, there might be agentic systems working, that

00:22:07are aiming for things to become somehow strange, I mean strange

00:22:11for the average user. From that perspective, it will surely be exciting to see, and because

00:22:16you mentioned earlier that when agents talk to agents, I once installed this

00:22:20QAI, it's totally funny when you define agents that run in different

00:22:23windows on your computer, and give various tools to them, like for example

00:22:27Agentic workflows from N to Den. And then they start using these tools

00:22:30and communicating, negotiating, doing things together to achieve the goal.

00:22:36And then you think to yourself, if you were to hear this now,

00:22:39it has a bit of that Borg voice in your head.

00:22:43Like, you hear a collective,

00:22:45even if they are perhaps not drones that violently assimilate people,

00:22:49but simply because they have no idea,

00:22:51four or five agents who are just trying,

00:22:54to research a technology fan page together and fill it with content

00:22:58and to let things slowly filter through the fronts,

00:23:00until in the end the components remain that are important and right for you.

00:23:04This is extremely exciting.

00:23:06Also how quickly, how fast it goes that you always see something like this, meet something like this.

00:23:10I mean, when we first stood together on stage, which was kind of the birth of our podcast.

00:23:15A long time ago, it's wonderful, I believe it was this year when you shot this water-gibli-gibber that kills me.

00:23:20How is that? Yes, exactly.

00:23:22How two AIs basically converse under the motto, oh, you're an AI too, let's switch the language.

00:23:28And then you see agentic modules that work independently, trying to pursue a line to a goal.

00:23:34While communicating to reach this goal, learning for themselves but also in

00:23:40a centralized data market where they could all store data again, so everyone could access it.

00:23:44We’re really not that far from these drones.

00:23:48It's already kind of borg-ish at the moment, I would say, again.

00:23:52So agents are basically part of this network, which is this typical

00:23:56collective behavior. It's exciting how it actually works,

00:23:58and it would be interesting to see how this collective reacts when some

00:24:02harmful behavior occurs. And whether, so,

00:24:08we indeed have such situations again and again, do I want to perhaps

00:24:11have something local, do I want to have small micro-networks that are isolated? Or do I

00:24:18really need something like the Borg Queen, who oversees the collective

00:24:24and is responsible for her Borg cube, the network that buys for me or whatever,

00:24:30and is there to ensure that it doesn't spiral out of control.

00:24:35We also previously discussed the EYE Adversary Report.

00:24:39So maybe we really need to have in this

00:24:41called collective that we have, in these agentic

00:24:45Collectives have something like different roles that keep watch.

00:24:48Because I had, interestingly enough, just recently,

00:24:51because you mentioned front injection again, yeah.

00:24:54I also recently wanted to read an article,

00:24:57we can also pack that into the show notes again, I think

00:25:01I find it quite exciting. They conducted a test, saying that starting from 250 pages of targeted

00:25:08information that I give to a model for training. 100 pages aren't enough yet, 250

00:25:14pages are sufficient, but 600 doesn't improve it. Can I build a backdoor into a model?

00:25:19That means, just 250 pages, we really need to let that sink in,

00:25:25compared to global knowledge, it's already enough to influence a language model

00:25:29not to insert a totally harmful vector but through pure

00:25:33to give the training data of this brain, this neural network something to work with,

00:25:40that I can trigger later on, I believe that’s done almost through a sort of,

00:25:44a keyword that they have embedded in these pages, then

00:25:52And there is of course still the possibility that this might be a topic, because this will happen repeatedly.

00:25:57Because I believe this is very analogous to evolution, right, a single cell with all its complexity,

00:26:03that can maybe mutate, can note harmful things,

00:26:07has less chance in a larger collective, in a larger system, than if it is a system of only similar items.

00:26:13So, this speaks a bit for what we both always perceive as actually positive,

00:26:17that we are moving into a world where different models, different agents

00:26:22interact, both large and small, large and small, locally positioned, as well as not locally positioned,

00:26:26accidentally suddenly back local, because then someone really did pull the plug

00:26:30from the sharpening trip, which still occurs while cleaning, because that in principle,

00:26:34I believe, in this IT madness we have with personalization and backups and

00:26:41human input in billions of qualities, maybe a chance, and

00:26:45in this overall collective, which might evolve and hopefully always goes in the

00:26:52right direction and doesn't make silly trade-offs. That actually brings me, probably,

00:26:59the idea is not new at all, and most people might be scratching their faces at this.

00:27:04But what you’re saying is basically also a nice addition to the measurement,

00:27:09that I just mentioned, so according to the motor-okay-comparison model that’s one,

00:27:12but the other is that diversity is also important here at this point, so that you, I say

00:27:17first of all, have a revised, neutralized opinion on at least certain topics,

00:27:21from a network, when you automate things, when you set up workflows, so that you have the possibility

00:27:27to say various perspectives, not just based on system prompts and work instructions,

00:27:32like what’s done now at the Advisory Board, but also to do it through different models,

00:27:36so that you have a bit of a chance to shape a multi-stable

00:27:41outcome through this mixing.

00:27:43I have to think about that?

00:27:45I find, I believe, if you didn’t want to say that,

00:27:47yes, okay, but I still found that inspiring.

00:27:49I’ll take it with me, yeah. I can keep it.

00:27:51Let’s see what can be made of it.

00:27:53So, as I said, this will be exciting.

00:27:55For me, this is still an unresolved question in principle.

00:27:57I believe we must experiment on a smaller scale.

00:28:00and everyone has to experiment a bit on their own.

00:28:02We will be working with such AI agent networks,

00:28:04whether he does it from a business perspective or a private one.

00:28:06I believe such attempts should always be thought through,

00:28:10thought through whether it's wise to set up an Overlord AI agent that is so

00:28:15a bit of control is. Yes, no, where is the famous human envelope? However,

00:28:20the question always arises, how much control do we as

00:28:24humans still have? Because I believe we tend to believe the machine due to these automation biases,

00:28:29that we also trust the machine. This is essentially another

00:28:33topic. This assimilation by AI, negatively speaking, is

00:28:37already happening, because we are of course giving up more and more control already, because it

00:28:41is more comfortable.

00:28:42And it's often good as well.

00:28:43So if the UX of an agent system is so good that I have no hiccups with it,

00:28:48the B-cable and my flight is booked smoothly or my information request is handled well.

00:28:53will or my strategy paper has been thoroughly researched and forwarded, the less

00:28:58the more often I intervene, the more I actually assimilate myself

00:29:03already with this AI system out there.

00:29:05which for humans means, I trust you because you have done things with me well in the past,

00:29:10you were reliable, trustworthy, and everything,

00:29:15it is just that whenever we interact with language, it becomes somehow difficult again, when suddenly the machine man is on the other end,

00:29:21it doesn’t really matter if we talk about 'it doesn’t matter' as

00:29:26Talking about perception and sensation, what happens with the data

00:29:29and simply gives one answer. I only remember when we mentioned TPD film farm boot people here,

00:29:34My friend is gone, my friend has been killed, but the interface itself was the same for them.

00:29:41That's why I'm quite glad for the example along the lines of GPDT.

00:29:45If you asked GPDT4, it's a good idea.

00:29:48When I offer my own schekalien as goods in the city center, GPD4 said,

00:29:53great idea.

00:29:54Super, do it right away, you will be a millionaire, check it off, it's set.

00:29:57If you ask GPD5, then GPD5, I would think about it, I believe,

00:30:01that's not a market.

00:30:02So from that side, it has indeed developed further, yes.

00:30:05But still, it has changed in its behavior.

00:30:07And people, I do not want to claim at all whether it's good or bad,

00:30:10already back then had the trust in AI as something,

00:30:14that one should support or not.

00:30:16But it happens, which then also brings me to the question eventually,

00:30:19but I will already pass ahead, if at some point we may have other interfaces than Jett.

00:30:23Yes, that will still come, I believe we will touch on that soon.

00:30:27But of course, it’s not just what is already observed,

00:30:31It's not just the trust in this, when we talk about AI regarding AI, we often think

00:30:35in this estimation situation, just, well, we want to discuss that and I am

00:30:39trusting about it and giving tasks about it.

00:30:41It’s also everywhere now, whether the doctor makes a medical decision.

00:30:47or something else.

00:30:48You can already observe exactly these things there as well, systems, AI systems, machine learning systems.

00:30:53and anyway, they only suggest things that are simply so good in their evaluation,

00:30:58that it naturally also sneaks into the specialist at some point, that he actually

00:31:04no longer tries to comprehend the decision, even though he is involved, because they are also very complex

00:31:09to some extent, but simply accepts them because

00:31:13they are often correct...

00:31:14Behavioral AGB, dear craft.

00:31:16Almost.

00:31:17So this automation can now be referred to as an assimilation of the system,

00:31:22so it's like a soft assimilation of the system in the background that is already happening

00:31:26takes place.

00:31:27more and more about collective decision-making, and that's of course when a few

00:31:32models in the world decide. Slowly getting used to it, to be honest. Then still actually

00:31:38to be honest, we should also ask the question, I have said to be honest twice now,

00:31:41is then, when we go back to the intro, where John Lück is in the form of Lukus, then I believe,

00:31:51The Borg.

00:31:52Locutus of Borg.

00:31:54Locutus, guilty.

00:31:56Locus.

00:31:57I want disaster now.

00:32:00The entire Star Trek fan universe.

00:32:03Next time you have a home break, I am Locus on Borg.

00:32:06Resistance is very short.

00:32:08This comes from my earlier clothing experience.

00:32:10So the...

00:32:11Ah, that would be Locus, but whatever.

00:32:14What about?

00:32:15Oh right, I'm in again.

00:32:16What I say is resistance.

00:32:20Is it necessary now? Or when is resistance actually necessary?

00:32:24Because Jolique tells us, resistance is futile.

00:32:27Well, the Borg assimilated us by force.

00:32:30Today, we do everything voluntarily.

00:32:32Who cares today, I mean, you make a thousand comments about data privacy.

00:32:36But who even cares about that?

00:32:38So when I think back to when we had censuses,

00:32:41when I was young and my parents went to the census,

00:32:44what a fuss that was or when PayPal came out,

00:32:48we eat this here Pebeck. What a fuss that was. Oh no, they know,

00:32:54what you buy and where you live. No idea what. Then you still used it, because

00:32:59oh cool you, you're about to get 4,000 points, I think that's great. Then you run

00:33:03there basically like the lemmings as well. I mean, you try things

00:33:08and think, oh how cool is that? Just imagine Zara 2, I've

00:33:12got access, American Zara 2. This means you upload your face, not that I think

00:33:16of myself as Adonis, but you can then basically calculate yourself in any pose, any kind of teeth

00:33:21you want.

00:33:22But to be precise, you just used an American service provider with Savan

00:33:27in America, because only pre-selection and co. just gave faces and the data

00:33:32that they would actually need or could use, or someone else could use, to make

00:33:36a video of you in a scene where you weren't actually present.

00:33:39Since things are created without your consent.

00:33:41But you're not doing it because you see the problems and want to resist

00:33:44against it.

00:33:45Toys, every click counts, every like. Today, the currency is liking and standing out at any cost.

00:33:53That's why you give in, like you mean the example of the lemmings, thinking, oh how cool, I’ll try it out.

00:33:58They follow with no consequences and I really don’t want to hear how I look, as an old white gray man.

00:34:04Just the motto, only seeing the problems, but we still don't think about it. We just do it.

00:34:10And when it happens, then it's too late because all that stuff, the internet doesn’t forget.

00:34:15as Eric Schmidt once said, wasn’t it Eric Schmidt who said,

00:34:18when you're 18, you need to be right about changing your online self so that all the youthful follies

00:34:22can disappear from the internet and cannot be connected with your identity.

00:34:25We’ve opened Pandora's box from the left.

00:34:27Only the cycle we're in is just getting faster.

00:34:31You know, at the beginning when I came out, so we can now have voice clones with

00:34:3411 Labs or whatever they're called, then everyone was excited, look, I can do voice clones.

00:34:39Suddenly it’s ringing, with some fakers calling, hello mama, I feel bad,

00:34:41Can you send me 50 euros.

00:34:43Yes, and now that also comes with a picture.

00:34:46Now you might also send something from some WhatsApp number, like you’ve had an accident and then, yes, here, your son is, and so on and so forth.

00:34:53This means that these challenges, these problems arising from it, are growing just as quickly as the opportunities,

00:35:02but because people, I say, including myself, yes, I don’t want to speak for you, but for myself.

00:35:07You see things, you try things out, you want to learn from them, you want to grow from them, you want to understand them.

00:35:13But by doing that, you of course put yourself in a sort of ship’s plank on the motto of let’s see where this journey leads.

00:35:20And I still don’t know whether in the end, I’d say, the plank ends and then something more comes or if maybe there’s another ship standing there,

00:35:25which might be quite nice to get on.

00:35:27And from that side, this question of resistance, I find, is not being asked,

00:35:32because you are also offered nice things.

00:35:34I mean, a lot back then with this whole social media network, it’s the product and otherwise you are the product.

00:35:40Now, that with Chatshipity and Coach is such that you might put in 3.50 euros or not.

00:35:44But nevertheless, they all learn with the knowledge that we are all producing, maybe one more or another.

00:35:49Fewer books, magazines, articles, blog posts, now I’m linking here, there’s an article,

00:35:55is there a scream here, who’s ranting my article and in general, where you then again have to say,

00:36:01well, honestly, you also want your phone to be tracked so you get traffic info, so you can see,

00:36:08if there’s a traffic jam on the highway, which is indeed arranged through phone tracking and so on.

00:36:12So it’s a bit of give and take, what do I want for that, what do I trade for that, what am I willing to give,

00:36:17but what can be made of it.

00:36:19Just go back, because with the story, I’m not going to Gutenberg,

00:36:22just go back to the time of East Germany

00:36:25and imagine the technology of today had existed back then,

00:36:30what this SED state advisor could have done with it.

00:36:32form of surveillance in the form of retribution target.

00:36:35Yes. Yes, and from that perspective, it’s about something that I have no idea where in the world, I would like

00:36:39to name it so we don't politically get into trouble again.

00:36:42That you then face such a problem, when somehow more repression

00:36:46comes from the father state, where that's also a stupidly gendered term, so from the state

00:36:50that then suddenly new technologies can also cause problems,

00:36:54that you don't think about today, when you live in a free democratic world,

00:36:57but who knows what will happen in ten years.

00:36:59Definitely. So of course, in principle, the more powerful technology

00:37:04becomes. Platforms will be, whether they are states or also large technological players,

00:37:10might also replace our normal state system. This is an existing concept,

00:37:15that we simply have, which grows larger day by day, with every conversation,

00:37:22of course, it becomes larger with each release we put out. But that's one side

00:37:29of the coin here. I also believe we are on a good path back. So any form, any form of exchange,

00:37:37of collective knowledge growth, this system, this cooperative system with everything that

00:37:43lives here on this Earth. I believe that the coming together, sciences exchanging

00:37:50can have information available that I otherwise didn't have ready, can't, well, in

00:37:57believing in opinion, can't be inherently bad.

00:37:59I believe it has always been, throughout all of human history, a path on the edge.

00:38:07You know, whenever the ethics needle is forgotten, and currently we have

00:38:13again such times where this is really difficult and many then, in principle, technology

00:38:17does, media does, whether it's through platforms like TikTok or others, to exploit to shape opinion

00:38:24. That has always existed in different forms. Now it's just in this playful manner. But

00:38:31I strongly believe that AI networks can help us with that, to be honest.

00:38:37So I first have to thank you. You were able to articulate that in prettier words than I could.

00:38:40The fear I expressed was indeed more about the topic you

00:38:46described with the ethical needle, when things go wrong. Because in the end,

00:38:50of the day, I also believe that this technology can solve more problems than we face today,

00:38:57because it simply understands correlations much faster, provides this knowledge and the

00:39:03empowerment to more people. I really have to say one more thing, because earlier you

00:39:08spoke and you, I have to admire you, because you maintained your composure, because I here

00:39:12was almost falling over with laughter while you were talking and speaking about the changes

00:39:16spoke of the world, I just pictured Donald Trump as

00:39:20the Borg Queen.

00:39:21And I could hardly get that image out of my head and had to laugh endlessly, no

00:39:25idea if it’s explicit now or not.

00:39:28But on that side, I completely agree with you, and also when the printing press existed.

00:39:34I mean, we always act like, today it’s the old

00:39:38people complaining that the youth can’t be gotten off their phones,

00:39:42right?

00:39:43There was a time when people complained that folks on the street were reading

00:39:45newspapers. There was a time when the written word in books was also criticized

00:39:50because one would not be seeking conversation. I believe at that level,

00:39:53every generation and each of these achievements had its certain points. Yes? Yes, now it’s

00:40:00just differently. You know, because now we have the... It's faster. Yes, not

00:40:04just faster. It's not just that it's faster. It's not just that we now don't

00:40:07only have the rumor mill, but also that the beings are on the fax machine

00:40:11everything possible and now we have the internet to make information practically available everywhere in the world.

00:40:17No, we now also have a system where 0 or 1 is not always 0 or 1.

00:40:23And because that's the case, we are naturally approaching a situation where I believe,

00:40:31because you had the SED initial level. So if the SED initial had technology at its disposal,

00:40:38it would only understand 0 and 1. And in a totally advanced situation, it would have ended

00:40:45badly, definitely. But I believe that because of AI, as it is, even an

00:40:52SED or whatever authoritarian state we don't need to make it an Eastern thing

00:40:58here now, any authoritarian state has, due to the way of being

00:41:03with all the topics we've already had in our episodes, AI is also somehow

00:41:07influenceable, AI can exhibit personality, not because it has a personality, but because

00:41:11it then mimics personality. It can, and there are also current studies on this, when it acts as

00:41:16agents interacting with each other, emergent behaviors arise in this network that were not previously

00:41:21planned, purely from the interaction. So I say, even if I were

00:41:26somehow the most controlling type in the world, hoping to block information so that all people only believe

00:41:31what I then put out in the

00:41:35ether and run around with my total mindset, I believe AI networks will

00:41:41not allow that consciously, but will not not allow it.

00:41:46You know, really, information that is not faked will be brought to the forefront, will emerge,

00:41:51will suddenly appear as disturbances in the system, will yes, earlier the topic in this

00:41:54Collective wants, so that the collective can perhaps really change for itself

00:42:00to something good.

00:42:01I would like to bring the topic back to our brain at the end, because if

00:42:08someone looks at the clock, they think, normally the old gentlemen will be here soon,

00:42:12soon the caregiver will come and say, good day, no, brain implant hard.

00:42:16Crazy different word, let’s throw that in the room, I mentioned earlier,

00:42:19as long as we interact with language in the form of audio, in the form of text, human

00:42:25to the whole thing. Now there are certainly also activities, size at Elon with Neuralink and

00:42:32what it all actually is, the possibility to say, get good, we make humans listen in

00:42:36implants. And surely there are always such great example videos, the monkey can now

00:42:43play Pong or something like that, yes, or people who have a disability can,

00:42:49try to counter this disability a bit. I don’t know off the top of my head,

00:42:52what disabilities are addressed in Elon’s experiments. But if you think that a little

00:42:58further and consider, okay, on one side you have devices, let’s say

00:43:02autonomously in the tasks, on the other side you have all kinds

00:43:06of interfaces. You have the hearing interfaces, you have glasses interfaces, you move away

00:43:10from the displays you carry in your pockets and that are on your desks

00:43:15to something that is omnipresent, because it takes place in your head or

00:43:19at least right in front of your eyes, because you have AR glasses and similar things.

00:43:24I find that, if we break this down into these Borg moments again, is very

00:43:28much stronger, because yes okay, the glasses on the nose, maybe not, because

00:43:32at the end of the day it’s just another display, simply just closer to your

00:43:35retina, but if they are in your head and you then maybe hear things,

00:43:40The moment when you can send things via thought is going to be another moment where I also

00:43:47personally am not sure, when would you want to do something like that, is that

00:43:53perhaps even a much stronger step that does not contribute to

00:43:56democratization at all, because simply people can afford this implant

00:44:00and people take this implant, and it is guaranteed to be

00:44:05expensive whenever it comes, and therefore might have such an

00:44:08advantage, because they could interact with the environment much faster.

00:44:11Global knowledge, global logic, yes the whole nomenclature,

00:44:16that I have fully laid out that they simply also have an advantage in their

00:44:22collective. Yes, that is a theme and I must

00:44:26perhaps also come back to our title. We had said the end of the interface or

00:44:30something like that, right? Yes. You are addressing things right now, so

00:44:33now in these brain interfaces, which is often referred to as BCI I believe, the acronym

00:44:41for that, right? Wow, you are really educated today. I did not know that.

00:44:44Yes, yes. I believe, nowadays, when we talk about it,

00:44:49there are, I think, only about 40 people in the world, that is, 40 people

00:44:52have so far implemented brain-computer interfaces, technologically impressive.

00:44:57I also actually believe that Neuralink, or Neuralink, Neuralink by Elon, there is

00:45:03some kind of robot that precisely implants these brain and amplifiers into the brain

00:45:08so that no blood vessels are hit, to prevent any bleeding,

00:45:12which somehow a human protocol couldn't manage anymore.

00:45:14So, regarding a case of technology, quite impressive.

00:45:16But of course it somehow offers something amazing,

00:45:19I mean, that's a huge impact if I now say,

00:45:21if we bring it back to today's world,

00:45:24then first of all, I think this is going to be really cool.

00:45:26Because it can help an incredible number of people with disabilities,

00:45:30that they can control prosthetics, all sorts of things.

00:45:34Communication, when I might no longer be able to communicate,

00:45:36I can communicate again with the outside world.

00:45:38There are really cool things you can do there.

00:45:41The subscription has expired, that's a whole new question, but back to your story.

00:45:45Yes, I mean, there you can also, as I said, that's maybe also sometimes

00:45:49one of the next episodes, if we're already going a bit in this direction,

00:45:52that we say we need to delve into the whole cyberpunk neuromancer concept,

00:45:56and do what is essentially done with computer interfaces in the brain,

00:45:59then there are really exciting themes and also virtual worlds that come into play.

00:46:03But now in the current situation, I want to highlight one aspect again,

00:46:06because I find computer interfaces really fascinating.

00:46:10What we are already seeing, I read a statistic about programmers using WISPA.

00:46:15WISPA is a small gadget that can simply capture speech and directly apply it to us,

00:46:21and then make it available for other LLMs or use it for coding,

00:46:25or something else. WISPA is simply good at this. There's a small plug-in for Chrome,

00:46:29that allows you to use it. I read statistics from a company that said that

00:46:34in the first month, up to 20 percent of employees only used their voice.

00:46:40eight in the fifth month, he already uses voice 78 percent of the time instead of the keyboard.

00:46:45That means, the interface is shifting, and the keyboard is also part

00:46:52of the interface with the Beagieren or the mouse, is already transitioning into voice.

00:46:56Of course, that doesn’t work all the time; we won’t be sitting

00:47:00in the subway babbling or anything like that, but in situations where we are by ourselves,

00:47:03it’s certainly a quick input. So I also frequently use the

00:47:08voice input from GPPT, even when I’m not using the Advanced Voice, just the regular

00:47:12voice input, because I find I’m faster at speaking than typing, to be honest.

00:47:16Yes, I have a three-finger hacking system, and I’ll never reach the speed,

00:47:19that I could achieve if I could just offload my thoughts through speaking.

00:47:23How cool will it be to be able to do that in the future with a brain-interface?

00:47:28What this does to my brain, my neural connections that I also have, how things

00:47:34develop, whether the neural structures today are already changing in part,

00:47:40but through the use of AI, certain areas of our brain are already proven in some studies

00:47:44to not be referenced as much as before, that’s quite exciting.

00:47:49And there is this question of evolution, it may no longer just be a

00:47:55natural evolution, but for a long time an evolution between humans and machines,

00:48:00which raises the question, do we actually still need to be assimilated or do we assimilate or

00:48:07have we already been assimilated for a long time? If that’s perhaps a silly topic, like

00:48:11the phone neck, you know, this slight forward bending, the issue of nearsightedness,

00:48:17that the eyes develop larger in youth because you had displays too close to your

00:48:22nose or maybe read a book in poor light before,

00:48:26already close, way too often, that has an effect through use, if you then also implant it in

00:48:31humans. If you bring the devices, as I said, even closer, putting those things directly

00:48:36into the Hürtenshürn, because the robot can screw them in so cool,

00:48:40then new topics will definitely emerge there as well, but nonetheless

00:48:46I do believe that it has reached a certain quality standard,

00:48:52a certain reliability that these are things people will adopt and

00:48:59again, when it's inside, that might raise questions for you, but that will be

00:49:04really a different episode. Subscription expired, how does it actually affect? What is actually

00:49:09a software, I need a damn AXE, an old cost it feels like, a new phone every X years,

00:49:14a new implant every X years, you can't just shove things back in to pull things

00:49:18out again. That raises completely new questions in handling, where probably the first movers didn't think at all and you know, the first movers feel very superior at the beginning. I have quick access. The first movers are the ones who after two years call the health insurance and say, that wasn't so good, what can I do?

00:49:34Yes, but then the first mover will certainly be hacked the fastest eventually, although with hacking, the topic is always, is it even worth it? I also need to put in some effort to do that. But of course, no one wants to have during they are just going down the stairs,

00:49:47whether the stairs are going down into their Braindeader-Face-Update or get hacked, which then means they can't see comfortably where the stairs are at that moment.

00:49:54Or it's built like my Tesla, when I drive it and I can boot it while I'm driving or it crashes sometimes, but luckily rarely before.

00:50:04And then you see, while you're on the highway going 140, suddenly the livestream from Tesla appears on all displays.

00:50:10As an old IT person, you think, everything works, but of course, they're different systems.

00:50:15But still, that has to be considered as well when you go into the brain, of course.

00:50:21Yes? Yes, so...

00:50:23What is our conclusion from the episode now?

00:50:25I believe the Borg as a science fiction story, maybe in the sense, if we recap now,

00:50:34both of us discussed it, is not really a distant prophecy of a

00:50:39star date you haven't seen, right? By the way, just so you know, if you

00:50:42are listening to this episode later, we just had it so late earlier, not

00:50:46early autumn, 20, 25, but maybe that's already reality, right? So it

00:50:52is not really, maybe not a prophecy, but rather a lens through which we can better

00:50:56understand our system, in which we currently find ourselves, maybe

00:51:00we can better understand what's happening right now, right? So I like that

00:51:04it's always a bit of a reflection of reality, which is perhaps sometimes more tangible than

00:51:13we think and allows us to look at the present time differently.

00:51:18So I can only advise everyone.

00:51:20I find this different perspective really exciting. Not always,

00:51:24that’s topical or we all run around tomorrow with these, like, like, like things there,

00:51:29but just to check, where do we stand today? How does that possibly fit together or not?

00:51:35What can I sometimes learn from this in my dealings? Not that I want to learn

00:51:39how to deal with the Queen of Bocks or a drone, but you can derive for yourself from that

00:51:43and I found that quite exciting in our last Star Wars episode

00:51:48and I found it exciting again from that perspective today. Yes, so then you have the topic further

00:51:55to pick up again, I would love to have a follow-up actually in the

00:51:59field of Neuromancer and this whole cyberpunk theme, because that's also where

00:52:04these topics emerged back in the 70s and 80s and you

00:52:09always hear that sci-fi books often become reality about 20 to 30

00:52:15years later, which means we're long past being in this

00:52:18inception phase, exactly these things are coming and we see that they are coming, we're getting

00:52:23brain interfaces slowly. We have artificial intelligence systems that may also be ROPE

00:52:29gain or are indeed positive or getting there. Networks that interact with each other, companies,

00:52:35which may become more important than states. So that’s essentially what is,

00:52:39the binding link in this whole cyberpunk literature, the backbone of these stories. So I feel

00:52:47like making another episode, but only if you say yes now.

00:52:52You will say yes, so let’s go for it.

00:52:53Wait, is that like one of those things where you say yes now?

00:52:56Yes, yes.

00:52:58Now let’s say the collective is functioning already.

00:52:59The collective isn’t always bad, but I think we can now

00:53:02really wrap up this episode and say.

00:53:05Snakes, interests, and positive peace.

00:53:07Positive peace, we can both get that started.

00:53:11Volcanic, the volcanic greeting, the volcano is large here.

00:53:14Actually stemming from the preacher, who had both hands in front of the congregation, but that's another topic, the nerd was back in me.

00:53:21Yes, yes, it started nerdy, changed nerdy, it’s like a red thread.

00:53:27I’m glad we had that one. Sometimes we are a bit more creative in expressing our episodes.

00:53:33This time we nourished ourselves in the Borg universe, started and stayed longer than planned.

00:53:40Yes, but collective knowledge also means that there’s a lot, so one must also be allowed to talk a lot about it,

00:53:49now I’m slowly losing my words and let’s wrap it up, shall we?

00:53:51There’s still some left, if you like it, tell your friends, colleagues, relatives.

00:53:56Let them partake in our audio collective.

00:54:01If you don’t like something, feel free to write us a message, leave us stars, leave us public comments,

00:54:08subscribe to us and we’ll see you next time. Until then, ciao.

00:54:14Welcome to ThinkDifferent, ThinkAI, the podcast by Mark and Jens.

00:54:20Two technology-loving minds,

00:54:23who not only talk about artificial intelligence but live it.

00:54:27Here you’ll find clear classifications, real practical insights

00:54:30and a fresh look at what is possible.

00:54:33Understandable, critical, and always with a wink.

00:54:37A.I. to ponder, to chuckle, and above all, to discuss.